University of British Columbia
Numerical methods in quantum chemistry to accelerate SCF convergence and calculate partial atomic charges
Abstract
dc:descriptionIn this thesis, we devise a series of minimization based methods to reduce the numbers of iterations needed to achieve convergence in SFC calculations. These methods are based on building linear combinations of Fock and Density matrices from previous iterations. Our techniques help to converge some systems for which established methods like DIIS and ADIIS fail. We also propose a scheme that combines our methods with DIIS. For some systems, this scheme reduces the number of iterations needed to reach convergence by more than 90\%. We also propose a method for assigning partial atomic charges in molecules. This method requires the division of the covalent part of the molecular density in partitions that take into account the ionic densities, the atomic number and the number of core/valence electrons in each atom. For small basis sets, our method provides atomic charges that are very similar in quality to those of natural population analysis, but with a lesser computational cost and a much more straightforward implementation. Finally, we collaborated with Dr. Ray Anderson in the identification of the absolute configuration of a series of chiral organic compounds. This was achieved by comparing the experimental electronic circular dichroism spectra with a theoretical one calculated with time-dependent density functional theory.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy - PhD
- Level thesis:degree_level
- doctoral
- Discipline thesis:degree_discipline
- Chemistry
- Grantor dc:publisher
- University of British Columbia
- Year dc:date
- 2018
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Garcia Chavez, Miguel Angel
Rights
dc:rights- Statement dc:rights
-
- Attribution-NonCommercial-NoDerivatives 4.0 International
- Language dc:language
- eng
Identifiers
dc:identifier.*- Handle dc:identifier
- http://hdl.handle.net/2429/67609